| { |
| "best_metric": 0.4033324209850535, |
| "best_model_checkpoint": "distilbert-base-uncased-finetuned-cola/run-3/checkpoint-804", |
| "epoch": 3.0, |
| "eval_steps": 500, |
| "global_step": 804, |
| "is_hyper_param_search": true, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 1.0, |
| "eval_loss": 0.5497495532035828, |
| "eval_matthews_correlation": 0.0928457264044978, |
| "eval_runtime": 0.7621, |
| "eval_samples_per_second": 1368.64, |
| "eval_steps_per_second": 86.606, |
| "step": 268 |
| }, |
| { |
| "epoch": 1.87, |
| "grad_norm": 3.906301498413086, |
| "learning_rate": 1.5090673992837595e-06, |
| "loss": 0.5538, |
| "step": 500 |
| }, |
| { |
| "epoch": 2.0, |
| "eval_loss": 0.5196466445922852, |
| "eval_matthews_correlation": 0.3636302567913777, |
| "eval_runtime": 1.0628, |
| "eval_samples_per_second": 981.371, |
| "eval_steps_per_second": 62.1, |
| "step": 536 |
| }, |
| { |
| "epoch": 3.0, |
| "eval_loss": 0.5164278745651245, |
| "eval_matthews_correlation": 0.4033324209850535, |
| "eval_runtime": 0.8583, |
| "eval_samples_per_second": 1215.213, |
| "eval_steps_per_second": 76.897, |
| "step": 804 |
| } |
| ], |
| "logging_steps": 500, |
| "max_steps": 804, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 3, |
| "save_steps": 500, |
| "total_flos": 96158136429468.0, |
| "train_batch_size": 32, |
| "trial_name": null, |
| "trial_params": { |
| "learning_rate": 3.9910861481057325e-06, |
| "num_train_epochs": 3, |
| "per_device_train_batch_size": 32, |
| "seed": 19 |
| } |
| } |
|
|